Revolutionizing the C3 AI Platform with Version 8
Table of Contents
- Introduction
- The Journey to Version 8
- Improving the Developer and Data Science Experience
- Low Code and Deep Code Experience
- Data Science Onboarding Experience
- Feature Store
- ML Pipelines
- Leveraging Existing Investments
- Conclusion
- FAQ
The Journey to Version 8
C3 has been on a journey to deliver a platform that will last the next 10 years. The goal was to build a platform that would provide a better low code, deep code experience for developers and architects, and a better data science experience for data scientists. The journey started with the delivery of B7, and when the team took a step back to think about what the next generation platform should look like, they wanted to make fundamental changes to the platform to make the developer and architect experience better.
The team started the project about four years ago, and they went through a very lengthy process of scoring and ranking the improvements that needed to be made. They picked a name for the project, Version 8, and started to put down what was on the top of their mind on a Confluence page. The improvements ranged from Incremental Type loading to improving performance, scalability, data ingest, and security.
Improving the Developer and Data Science Experience
Low Code and Deep Code Experience
The team at C3 wanted to provide a better low code, deep code experience for developers and architects. They wanted to make getting access to environments faster and easier, making provisioning time or deployment times faster. They wanted to address issues with UI bundling and make them available so that as the user experience is being edited, it's immediately available.
For data science teams, C3 wanted to make sure that they're providing the right experience for them. They received some very pointed feedback that they needed to provide a better data science onboarding experience. They wanted to make it easier for data scientists to onboard and use the platform. They also wanted to provide a better experience for developers and architects.
Data Science Onboarding Experience
C3 wanted to make it easier for data scientists to onboard and use the platform. They wanted to provide a python first experience with data sets. With data sets, data scientists can open up their Jupyter lab notebook and start using the knowledge they have to explore data, prepare data, develop their models, develop their features, and eventually get a production-ready app.
Feature Store
C3 wanted to provide a better feature store experience. They wanted to add support for time series and non-time series features. They wanted to provide a much cleaner, much more obvious view of ML features, the relationships to models, training data, and the ML subjects that are eventually trained and inferred on. They also wanted to persist the features so that there's more repeatability and improved performance.
ML Pipelines
C3 wanted to make ML pipelines simpler. They wanted to minimize the additional code that You need to write. They also wanted to make it easier to declare features in a pipeline themselves.
Leveraging Existing Investments
C3 wanted to make it easier to leverage existing investments. They wanted to make it easier to reuse data products across a portfolio of applications. They wanted to provide better compatibility with existing infrastructure.
Conclusion
C3 has been on a journey to deliver a platform that will last the next 10 years. They have made fundamental changes to the platform to make the developer and architect experience better. They have also made improvements to the data science onboarding experience, the feature store, and ML pipelines. They have made it easier to leverage existing investments.
FAQ
Q: What is the goal of Version 8?
A: The goal of Version 8 is to provide a better low code, deep code experience for developers and architects, and a better data science experience for data scientists.
Q: What improvements have been made to the platform?
A: The improvements range from incremental type loading to improving performance, scalability, data ingest, and security.
Q: What is the data science onboarding experience?
A: The data science onboarding experience is a python first experience with data sets. With data sets, data scientists can open up their Jupyter lab notebook and start using the knowledge they have to explore data, prepare data, develop their models, develop their features, and eventually get a production-ready app.
Q: What is the feature store?
A: The feature store is a much cleaner, much more obvious view of ML features, the relationships to models, training data, and the ML subjects that are eventually trained and inferred on.
Q: What improvements have been made to ML pipelines?
A: ML pipelines have been made simpler. The additional code that you need to write has been minimized. It is also easier to declare features in a pipeline themselves.
Q: What is the goal of leveraging existing investments?
A: The goal of leveraging existing investments is to make it easier to reuse data products across a portfolio of applications and to provide better compatibility with existing infrastructure.